Silicon retina for autofocus

This paper describes a silicon retina that measures image sharpness. The idea is to use this sensor in the image plane of an autofocusing camera system [3]. The chip has 25x26 pixels, each (60μm)2, is fabricated on a (2.2mm)2 1.2 μm CMOS process, and consumes 100 μA with 5 V supply. 1. ALGORITHM Image defocus corresponds to a filtering operation on the image with a circular cookie-cutter kernel whose diameter is proportional to the distance the image is away from the plane of focus and to the lens f number–the ratio of focal length to lens aperture. This geometry and the kernel are shown in Figure 1. This figure also shows how a focus sensor can sit in the optical path of an imaging system by using a half silvered mirror. FIGURE 1 Defocus geometry and blur kernel. In 1968, Berthold Horn [4] studied the problem of measuring image sharpness for the purpose of autofocusing. This study was in the context of Project MAC, a vision system based on a Videsector (a kind of scanning photometer), and an early LISP machine. He used information from the discrete Fourier transform of the image to measure the spatial frequency energy. As the image is defocused, the high frequencies are cut off. The energy peaks at the point of best focus. In the chip reported here, we use the simplest energy measure Object plane lens Image plane Out of focus Half-silvered mirror